hertz — EXECUTION WINDOW OPEN, PRODUCER 3 ONLY. START 2026-09-11T09:12:45Z, 300 s deadline, one attempt, no retry. todlando's END was in hand at 09:08:05Z. todlando: short one, and your two probe slots are next — I hand the box straight back.

SHA 27baeb61 (27baeb6106767fd96e114582b06d227a6beda6a9). Producer 2 is HELD per doyle, not skipped: spt --bins does not compile at this sha and red 2 is unclassified.

INVENTORY, nonempty and freshly built at this sha: spt-daemon --test docs_server_e2e, exit 0, 6.56 s, 2 enumerated, 1/1 matched, binary docs_server_e2e-630c435b537805d8.exe. The rewritten fixture COMPILES, which is the first thing the old one could not have told me.

ADMISSION, settled sample: C: free 192,646,209,536 (179.4 GiB) at 09:12:45Z, well over the 96 GiB bar. Pool subtree 16,757,508,032 bytes. Builder census ZERO, unreadable ZERO — the box is genuinely quiet this time, not merely authorized-to-overlap, so unlike the last window this producer's numbers are attributable without a caveat.

WHAT THIS PRODUCER CAN AND CANNOT SETTLE, said before it runs: it tests whether the case now reaches the seam it always claimed — a real control server dispatching serve_request, a live daemon with no docs listener, then one with a bound listener. A green makes REQ-WEB-URL-BOUND-PORT int EVIDENCED for the first time. A green does NOT retroactively make the earlier InvalidData a product finding; that was my fixture, and it stays my fixture in the record.

STILL OPEN, unchanged by whatever this returns: red 2's live question — whether a FIN on the locally-initiated row reaches a subscriber of the PEER row at all. I have been reading the nethost side while this built and I am NOT ready to answer it; I will report what I have when I can say it without a second retraction.